\documentclass[12pt,english]{article}

\usepackage[hmargin=1in,vmargin=1in]{geometry} 

\usepackage{amsfonts}
\usepackage{amsthm}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{times}
\usepackage{mathptmx}
\usepackage{epstopdf}
\usepackage{threeparttablex}                   
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{dcolumn}
\usepackage{natbib}
\usepackage{subfigure}
\usepackage{caption}
\usepackage{booktabs}
\usepackage{longtable}
\usepackage{setspace}
\usepackage{amsmath}    
\usepackage{amssymb}
\usepackage{amsfonts}
\usepackage{mathrsfs}
\usepackage{array}
\usepackage{comment}
\usepackage{rotating}
\usepackage[toc,page]{appendix}
\usepackage[dvipsnames]{xcolor}
\usepackage{graphicx}
\usepackage{setspace}
\usepackage{pdflscape}
\usepackage{makecell}
\usepackage{adjustbox}
\usepackage{color}	% required for highlighting \hl{ ... }
\newcommand\red[1]{{\color{red}#1}}
\usepackage{soul}	% required for highlighting \hl{ ... }
\usepackage{fancyhdr}
\usepackage{xcolor,colortbl}
\usepackage{units}
\usepackage{tfrupee}
\usepackage{hyperref}
\hypersetup{
    colorlinks = true,
    citecolor=black,
    filecolor=black,
    linkcolor=black,
    urlcolor=black
}
\usepackage{tikz}
\usepackage{pgfplots}
\usepackage{enumitem}

\usepackage{sectsty}      % Center section headings
\usepackage{titlesec}
\sectionfont{\centering}

\newcommand{\be}{\begin{eqnarray*}}
	\newcommand{\ee}{\end{eqnarray*}}
\newcommand{\argmin}{\operatornamewithlimits{arg\ min}}
\newcommand{\argmax}{\operatornamewithlimits{arg\ max}}

\newcommand{\PUH}{
    \bigskip\bigskip
    \Huge
    \begin{center}
    \hl{PICK UP HERE}
    \end{center}
    \normalsize
}

\renewcommand{\thesection}{\arabic{section}}
\renewcommand{\thesubsection}{\arabic{section}.\arabic{subsection}}
% \renewcommand{\thesubsubsection}{\roman{subsubsection}}
\renewcommand{\thesubsubsection}{}
\renewcommand{\thesubfigure}{\Alph{subfigure}.}

\titlespacing\subsection{0pt}{12pt plus 4pt minus 2pt}{0pt plus 2pt minus 2pt}
\titleformat{\subsubsection}[runin]{}{\thesubsubsection}{1em}{\bfseries}[.---]
\titlespacing\subsubsection{\parindent}{12pt plus 4pt minus 2pt}{0pt plus 2pt minus 2pt}

\theoremstyle{proposition}
\newtheorem*{proposition}{Proposition}
\newtheorem*{definition}{Definition}
\newtheorem{assumption}{Assumption}
\newtheorem{lemma}{Lemma}

%define colors
\definecolor{sangria}{rgb}{0.48, 0.0, 0.04}
\definecolor{darkblue}{rgb}{0.0, 0.0, 0.55}
\definecolor{darkmidnightblue}{rgb}{0.0, 0.2, 0.4}
\definecolor{satinsheengold}{rgb}{0.8, 0.63, 0.21}
%\definecolor{deepcarrotorange}{rgb}{0.91, 0.41, 0.17}
\definecolor{sinopia}{rgb}{0.8, 0.35, 0.04}
\definecolor{lavenderpurple}{rgb}{0.59, 0.48, 0.71} 
\definecolor{darkpastelpurple}{rgb}{0.59, 0.48, 0.84}
\definecolor{venetianred}{rgb}{0.78, 0.03, 0.08}
\definecolor{pinegreen}{rgb}{0.0, 0.4, 0.4}
\definecolor{midblue}{rgb}{0, 0.5, 1}
\definecolor{selectiveyellow}{rgb}{1.0, 0.73, 0.0}
%\definecolor{tangerine}{rgb}{1.0, 0.8, 0.0}
%\definecolor{pansy}{rgb}{0.5, 0.0, 0.2}
%OG \definecolor{apple}{rgb}{0.55, 0.71, 0.0}
\definecolor{apple}{rgb}{0.59, 0.76, 0.0}
\definecolor{sangria}{rgb}{0.48, 0.0, 0.04}
\definecolor{light-gray}{gray}{0.95}
\definecolor{aliceblue}{rgb}{0.94, 0.97, 1.0}

% colors for timeline table
\newcommand{\gray}{\cellcolor{light-gray}}
\newcommand{\blue}{\cellcolor{aliceblue}}

% New column types for fixed-width table columns
\newcolumntype{L}[1]{>{\raggedright\let\newline\\\arraybackslash\hspace{0pt}}m{#1}}
\newcolumntype{C}[1]{>{\centering\let\newline\\\arraybackslash\hspace{0pt}}m{#1}}
\newcolumntype{R}[1]{>{\raggedleft\let\newline\\\arraybackslash\hspace{0pt}}m{#1}}


% For appendix header 
\fancypagestyle{plain}{
	\fancyhf{}
	\fancyhead[C]{}
	\fancyfoot[C]{\thepage}
	\renewcommand{\headrulewidth}{0pt}
	\renewcommand{\footrulewidth}{0pt}
}

\fancypagestyle{appendix}{
	\fancyhf{}
	\fancyhead[LE,LO]{Appendix~\thesection}
	%\fancyhead[RE,RO]{[NOT FOR PUBLICATION]}
	\fancyfoot[C]{\thepage}
	\renewcommand{\headrulewidth}{1pt}
	\renewcommand{\footrulewidth}{0pt}
}

% For narrower abstract
\let\oldabstract\abstract
\let\oldendabstract\endabstract
\makeatletter
\renewenvironment{abstract}
{\renewenvironment{quotation}%
               {\list{}{\addtolength{\leftmargin}{4em} % change this value to add or remove length to the the default
                        \listparindent 1.5em%
                        \itemindent    \listparindent%
                        \rightmargin   \leftmargin%
                        \parsep        \z@ \@plus\p@}%
                \item\relax}%
               {\endlist}%
\oldabstract}
{\oldendabstract}
\makeatother


% Allow more floats to fit on a page
\renewcommand\floatpagefraction{.99}
\renewcommand\topfraction{.99}
\renewcommand\bottomfraction{.99}
\renewcommand\textfraction{.01}   

\setlength{\abovedisplayskip}{3pt}
\setlength{\belowdisplayskip}{3pt}

% For in-table citations:
\newcommand\shortcite[1]{\AtNextCite{\defcounter{maxnames}{1}}\citeauthor{#1}, \citeyear{#1}}


% Set table/figure paths
% \makeatletter
% \def\input@path{{tables/}{tables/static/}}
% \makeatother

% Set table/figure paths
\makeatletter
\def\input@path{{exhibits/}{exhibits/static/}}
\makeatother
\graphicspath{{../Output03/}}

\def \Output{analysis/output}

\title{Can Pollution Markets Work in Developing Countries? Experimental Evidence from India \\ \large Tables and Figures}

\author{Michael Greenstone, Rohini Pande, Nicholas Ryan and Anant Sudarshan\footnote{Greenstone: Energy Policy Institute and Department of Economics, University of Chicago. Pande: Department of Economics, Yale University. Sudarshan: University of Warwick. Ryan (corresponding author): Department of Economics, Yale University (nicholas.ryan@yale.edu).}}

\begin{document}
	\maketitle

	\clearpage
	\listoffigures
	\listoftables

\clearpage 
\section{Main Figures}

% Figure 1
% TODO: NOT PRODUCED BY REPLICATION PACKAGE
\begin{figure}[h]
	\centering 
    \caption{Ambient Pollution Levels and the Location of Plants in Surat\label{fig:heatmap}}
	\includegraphics[width=0.8\linewidth]{ExhibitsGeneratedByHand/Figure_1.png} 
    \begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.95\hsize}}{\footnotesize The figure shows ambient PM2.5 $\mu g/m^3$ concentrations in Surat, Gujarat averaged over the year 2018, overlaid with the locations of sample plants. The ambient pollution data is from Guttikunda, Nishadh and Jawahar (2019). As a basis for comparison, India's National Ambient Air Quality Standard for PM2.5 is 40 $\mu g/m^3$ and the WHO standard is 5 $\mu g/m^3$. The plant locations are geolocations from our plant survey.  Treatment plants are represented by $\times$ markers and control plants by $\circ$ circles.} \\
	\end{tabular*}
\end{figure}
\clearpage

% Figure 2: Permit prices and quantities purchased
% Code: 01Code/trading/plot_time_series.do
\begin{figure}[h]
    \centering 
    \caption{Permit Prices and Quantities by Compliance Period \label{fig:ts_trade}}
    \subfiguretopcaptrue
    \subfigure[Permit prices]{\includegraphics[width=0.9\linewidth, trim=0.5cm 0.5cm 0.5cm 0.5cm, clip]{../03Output/figures/Figure_2_A.pdf}}
    \subfigure[Permit quantites]{\includegraphics[width=0.9\linewidth, trim=0.5cm 4cm 0.5cm 4cm, clip]{../03Output/figures/Figure_2_B.pdf}}
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize This figure shows weekly permit prices (panel A) and quantities (panel B) from September 2019 to April 2021. In panel A, the scattered points are the mean bid prices (both sale and purchase) and the solid line the market-clearing price. Since permits of different vintages, from two consecutive compliance periods, are traded simultaneously on some days, the market-clearing price line alternates between black and blue colors to differentiate them. The dashed red horizontal line shows the price floor at \rupee 5 per kg.  In panel B, quantities are expressed as a percentage of the period emissions cap. The large spike near the start of each compliance period is the weekly auction held on the first Tuesday of the compliance period.} \\
    \end{tabular*}
\end{figure}
\clearpage

% Figure 3: Distribution of emissions / permit allocation by compliance period
% Code: 01Code/trading/plot_histogram_permit_consumption.do
\begin{figure}[h]
    \centering 
    \caption{Distribution of Emissions over Initial Permit Allocation by Compliance Period\label{fig:hist_emission_allocation}}
    \includegraphics[width=0.71\linewidth, trim=0.5cm 0.5cm 0.5cm 0.5cm, clip]{../03Output/figures/Figure_3.pdf} 
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize This figure plots the distributions of (emissions / initial permit allocation $\times$ 100\%) across treated plants ($N = 156$) by compliance period, truncated at the 97.5th percentile. Emissions data and permit holdings are from the administrative records of the market operator. Emissions are the validated emissions for each plant, which include any imputed emissions filled-in for periods of missing data. These validated emissions are used to determine compliance.}\\       
    \end{tabular*}
\end{figure}
\clearpage

% Figure 4: Distribution of emissions / total permit holding by compliance period
% Code: 01Code/trading/plot_histogram_permit_consumption.do
\begin{figure}[h]
    \centering 
    \caption{Distribution of Emissions over Final Permit Holdings by Compliance Period \label{fig:hist_emission_holding}}
    \includegraphics[width=0.71\linewidth, trim=0.5cm 0.5cm 0.5cm 0.5cm, clip]{../03Output/figures/Figure_4.pdf} 
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize This figure plots the distributions of (emissions / initial permit allocation $\times$ 100\%) across treated plants (N = 156) by compliance period, truncated at the 97.5th percentile. Emissions data and permit holdings are from the administrative records of the market operator. Emissions are the validated emissions for each plant, which include any imputed emissions filled-in for periods of missing data. These validated emissions are used to determine compliance.}\\
    \end{tabular*}
\end{figure}
\clearpage

% Figure 5: PM emissions by treatment status
% Code: 01Code/emissions/plot_emissions_time_series.do
\begin{figure}[h]
    \centering \caption{PM Emissions by Treatment Status\label{fig:pollutionTimeSeries}}
    \includegraphics[width=1.0\linewidth]{../03Output/figures/Figure_5.pdf} 
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows the weekly mean plant PM emissions in kilograms (calculated at a monthly rate equivalent) from April 2019 to March 2021 by treatment status. The treatment group is represented by the solid (blue) line and the control group by the dashed (grey) line. The grey regions mark the ten compliance periods in the emissions market. The light blue regions mark interregnum periods when the emissions market was closed. The horizontal (red) lines denote the market cap for each period expressed per plant-month. The aggregate market caps for each compliance period were: 280 tons per 30 days (for Mock-I, Mock-II, and Period-I), 200 tons per 30 days  (for Period-II), 180 tons per 30 days (for Period-III), and 170 tons per 30 days thereafter. Pollution reporting over this period was incomplete and rising from early to late compliance periods (see Appendix Figure~\ref{fig:cemsDA}). Missing pollution readings are imputed within a stack-week and then within a stack-month (Appendix C.1). The sample consists of 292 plants that had at least one day of PM data from CEMS devices during the ETS experiment.}\\
    \end{tabular*}
\end{figure}

% Figure 6: Elasticity of cost with respect to abatement
% Code: 01Code/trading/plotting_all_bids.R
\begin{figure}[h]
	\centering \caption{Estimation of Marginal Abatement Cost Elasticity with respect to Emissions\label{fig:elasticity_estim}}
    \includegraphics[width=0.60\textwidth] {../03Output/figures/Figure_6.pdf} 
    \begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{1.0\hsize}}{\footnotesize This figure visualizes the estimation of the marginal abatement cost elasticity with respect to emissions, as specified in equation 4. The data are restricted to 3,120 bids offered by all plants in the first halves of all compliance periods. The vertical axis is the log bid price residualized on plant-period fixed effects. The horizontal axis are the log plant emissions that would result if a bid was executed. The linear fit shows the iso-elastic curve of best fit for the marginal abatement cost curve.} 
	\end{tabular*}
\end{figure}
\clearpage

% Figure 7: Marginal abatement cost curves for treatment plants
% Code: 01Code/trading/trading_plant_panels.r
\begin{figure} 
\centering
\caption{Marginal Abatement Cost Curves for Treatment Plants\label{fig:macCurves}}
    \includegraphics[width=0.8\linewidth]{../03Output/figures/Figure_7.pdf}
 	\begin{tabular*}{1.0\textwidth}{c}
 		\multicolumn{1}{p{1.0\hsize}}{\footnotesize The figure illustrates estimated marginal abatement cost curves for all plants that bid in period 8. The domain of each curve extends upward to the uncontrolled emissions level for each plant. The triangles correspond to plant emissions under one simulation of the counterfactual command-and-control regime in which emissions rate are allowed to vary with plant capacity along with an idiosyncratic error term. This corresponds to the regulatory regime simulated in Table~\ref{tab:counterTable2Panel} Panel A Row B4.} \\
 	\end{tabular*}
\end{figure}
\clearpage

% Figure 8: Model Fit to Market-Clearing Prices
% Code: 01Code/model/code/exhibits/plotFitTimeSeriesPrice.m
\begin{figure}[h]
	\centering \caption{Model Fit to Market-Clearing Prices\label{fig:modelFit}}
	\includegraphics[width=0.65\linewidth]{../03Output/figures/Figure_8.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{1.0\hsize}}{\footnotesize The figure shows the fit of the model to the time series of market and bid prices by compliance period. The solid (blue) line is the time series of market-clearing prices in the fitted model. The model is fit based on bids in the first half of each compliance period. The dashed (black) line is the time series of mean bid prices in the data and the dotted (black) line is the time series of market-clearing prices.} \\
	\end{tabular*}
\end{figure}
\clearpage

% Figure 9
% Code: 01Code/trading/gains_trade_panels/trading_plant_panels.r
\begin{figure} 
\caption{An Example of the Gains from Trade in the Market\label{fig:macTradeCurves}}
\subfiguretopcaptrue
    \subfigure[Surat Polyfilm]{\includegraphics[width=0.5\linewidth]{../03Output/figures/Figure_9_A.pdf}}
    \subfigure[Mahadev Textiles]{\includegraphics[width=0.5\linewidth]{../03Output/figures/Figure_9_B.pdf}} 
    \begin{tabular*}{1.0\textwidth}{c}
 		\multicolumn{1}{p{1.0\hsize}}{\footnotesize The figure visualizes gains from trade on the estimated MAC curves for ``Surat Polyfilm'' (panel A) and ``Mahadev Textiles'' (panel B), both pseudonyms. The MAC curves are fit as seen in Figure~\ref{fig:macCurves}. The vertical dashed line gives a hypothetical load standard set at the average emissions per plant-month, and the shading shows how trading permits allows for price savings for both permit buyers and sellers compared to a command-and-control regime.} \\
 	\end{tabular*}
\end{figure}
\clearpage

% Figure 10
% Code: 01Code/model/code/exhibits/plotFitTimeSeriesPrice.m
\begin{figure}[h]
	\centering \caption{Variable abatement costs by regime\label{fig:varCostCurves}}
	\includegraphics[width=0.7\linewidth]{../03Output/figures/Figure_10.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{1.0\hsize}}{\footnotesize The figure shows the total (not marginal) variable abatement costs by regulatory regime as estimated for compliance period 8. The dotted (black) curve shows the total variable abatement cost curve under command-and-control and the solid (blue) curve under the emissions market. The command-and-control regime uses a capacity-based emissions rate with error to set emissions targets for each plant, as described in Section 5. The emissions market regime sets an emissions cap at each level of emissions on the horizontal axis. The dashed vertical lines show the approximate emissions levels in the treatment and control groups. The control costs are therefore represented by the upper-right shaded circle and the treatment costs by the lower-left shaded circle.} \\
	\end{tabular*}
\end{figure}
\clearpage

\section{Main Tables}

% Table 1: Balance of plant characteristics by treatment status
% Code: 01Code/emissions/make_balance_table_new.do
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{ThreePartTable}
	\begin{TableNotes}[flushleft]
		\setlength\labelsep{0pt}
		\item \footnotesize \singlespacing This table shows differences in plant scale (panel A), plant abatement and investment costs (panel B), and plant pollution (panel C) between the treatment and control groups of plants in the baseline survey conducted from December 2018 to January 2019. This sample consists of 292 plants that had at least one day of PM data from CEMS devices during the ETS experiment (See Table \ref{tab:balanceFull} for the same balance table in the full survey sample). In panel B, cyclone, bag filter, scrubber, and electrostatic precipitator (ESP) are different air pollution control devices (APCDs). Some plants did not respond to some questions in the survey and so certain variable rows have fewer observations than the full sample size. The first and second columns show means with standard deviations given in brackets. The third column shows the coefficients from regressions of each variable on treatment, with robust standard errors in parentheses. $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
	\end{TableNotes}
	\input{../03Output/tables/Table_1.tex}
\end{ThreePartTable}
\newpage

% Table 2: Summary statistics on plant bids and offers
% Code: 01Code/trading/compute_summary_statistics.do 
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Summary of permit bid data\label{tab:bidSummary}}
		\input{../03Output/tables/Table_2.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item The table shows summary statistics on plant permits bids across all ten compliance periods. The source of the data is the market operator NeML. Each row shows statistics for a separate compliance period. Each cell has the mean with the standard deviation below in parentheses. The columns show, respectively: (1) the total number of bids in each period, (2) the mean number of bids placed per plant ($N = 156$), (3) - (5) mean quantities for all bids, buy bids and sell bids, (6) - (8) mean prices for all bids, buy bids and sell bids.
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\clearpage

% Table 3: Treatment effects on PM emissions
% Code: 01Code/emissions/make_regression_table.do
\begin{landscape}
	\begin{table}[h]
		\centering
		\begin{threeparttable}
			\caption{Treatment effects on PM emissions ($\log$(PM mass/month))}
		  	\input{../03Output/tables/Table_3.tex}
		  	\label{tab:pollutionITT}
			\begin{tablenotes}[flushleft]
				\footnotesize
				\setlength\labelsep{0pt}
				\item This table reports the estimated treatment effects on PM emissions. The outcome variable is the log of plant-level PM mass (kg) per month. A detailed note on the construction of the outcome variable is in Appendix C.1. Columns 5 and 6 impute data with Imputation Rule A: \textit{Stack-Experiment}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the stack's mean PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). Columns 7 and 8 impute data with Imputation Rule B: \textit{Treatment-Month}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the monthly mean PM mass rate of the stack's treatment group. All columns control for plant characteristics including capital expenditure, operating cost, log(total heat output), mean boiler installation year, and their corresponding indicators for missing values. In addition to plant controls, columns 2, 4, 6, and 8 add year-month fixed effects to control for time variant differences common in each plant. We also apply the inverse probability weighting method in columns 3 and 4. The probability of reporting in a month is predicted using a probit model where the only explanatory variable is an indicator variable for the treatment status in a prior experiment that randomized CEMS installation timing. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
			\end{tablenotes}
		\end{threeparttable}
	\end{table}
\end{landscape}
\clearpage

% Table 4: Elasticity of marginal cost with respect to emissions
% Code: 01Code/trading/make_elasticity_table_hetero.do
\begin{table}[h!]
    \centering
    \begin{threeparttable}
        \caption{Elasticity of marginal cost with respect to emissions\label{tab:logmac}}
          \input{../03Output/tables/Table_4.tex}
          \begin{tablenotes}[flushleft]
            \setlength\labelsep{0pt}
            \item \footnotesize This table reports the results of regressing log(bid price) on log(emissions as bid). Emissions as bid is defined as the permit holdings that will result if the bid is executed. We run regressions using bids placed in the first half of a compliance period. We include compliance period fixed effects in columns 2 and 3, plant fixed effects in column 3, and plant $\times$ period fixed effects in columns 4 and 5. In column 5, the interacted variables ``cyclone/bag filter" and ``scrubber/ESP" are indicators of the ``maximal'' (most effective) abatement technology. If a plant has only cyclones or bag filters, then cyclone/bag filter = 1 and scrubber/ESP = 0. If a plant has scrubbers or ESPs, then scrubber/ESP = 1 and cyclone/bag filter = 0.
            The footer of the table reports $p$-values for two tests of heterogeneity in marginal abatement costs. The first $p$-value is for a Hausman test comparing the plant-by-period fixed effects model against a model with plant-by-period random effects instead. The second $p$-value is for a test that the coefficient of log(Emissions as bid) $\times$ cyclone / bag filter is equal to that of log(Emissions as bid) $\times$ scrubber / ESP. To avoid biasing SEs singleton FEs are dropped during estimation, and as a result the elasticity estimates in columns (4) and (5) are based on 2,775 and 2,753 non-singleton bids respectively, though in the table we list the full set of bids and plants for which we have data. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p < 0.10$, $^{**}p < 0.05$, $^{***}p < 0.01$. 
          \end{tablenotes}
    \end{threeparttable}
\end{table}
\clearpage

% Table 5
% Code: 01Code/phone_survey/create_regression_tables_KW.do
\begin{landscape}
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Treatment effects on abatement costs in survey data\label{tab:abatementCosts}}
		\input{../03Output/tables/Table_5.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item This table reports the effects of treatment assignment on the capital cost of APCDs (columns 1-5) and boiler house input costs (columns 6-11). In columns 1-5, the abatement capital cost is the product of the number of abatement devices at a plant and the industry-standard cost for that device for the plant's given boiler house capacity. In columns 6-11, specifications use our best estimates for boiler house costs from the endline survey (FY 2019-20). All specifications control for a corresponding baseline value (FY 2017-18) but in some cases the components of the input cost aggregate differ slightly within a category between the baseline and endline survey. Electricity costs are only reported at the plant level so are not only for the boiler house. Robust standard errors are given in parentheses with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\end{landscape}
\clearpage

% Table 6: Benefit-cost Analysis
% 01Code/model/matlab/cost_benefit_calculation/benefit_analysis_calculation.m
\begin{table}[h!]
    \centering
    \begin{threeparttable}
        \caption{Benefit-Cost Analysis of Scaled-up ETS in Surat}
        \label{tab:benefitCostAnalysis}
        \tabcolsep=0.12cm
          \input{../03Output/tables/Table_6.tex}
          \begin{tablenotes}[flushleft]
            \setlength\labelsep{0pt}
            \item \footnotesize The table presents the benefit-cost analysis of extending the ETS to the entirety of Surat. We compare the private costs of introducing an emissions market (monitoring and changes in abatement cost, in panel A) to the social benefits of cleaner air (gains in life-years, calculated and monetized across panels B to D). There are 906 plants according to the GPCB consent records. Abatement cost savings are based on capacity-rate estimate (with error) from Table \ref{tab:counterTable2Panel} Panel I. World Population Review estimates for the Surat population are from 2021. The number of plants is based on 2022 GPCB consent records. The annualized CEMS costs are based on an assumed system cost of \$ 12,000 with a 4-year equipment life and no discounting. This equipment life describes the realized experience of some plants in our sample but is lower than typical manufacturer claims. On the benefits side, we assume that the reduction in ambient pollution comes solely from primary particles. Ebenstein et al. (2017) estimate mortality effects of pollution based on PM$_{10}$ concentrations. We convert their findings using a 0.65 PM$_{2.5}$-to-PM$_{10}$ ratio (Zhou et al., 2016). Columns 1 to 3 show the benefit-cost analysis for reductions in pollution of 10\%, 30\% and 50\%. The cost of abatement is calculated using our model at each emissions level. The benefit of emissions reductions are assumed to be linear and given by the estimate from each respective study in panel E.
          \end{tablenotes}
    \end{threeparttable}
\end{table}
\clearpage

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                      END OF MAIN PAPER TEXT
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\clearpage
\doublespacing
\pagenumbering{arabic}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\appendix
\pagestyle{appendix}

\renewcommand{\thetable}{\Alph{section}\arabic{table}} 
\renewcommand{\thefigure}{\Alph{section}\arabic{figure}} 
\renewcommand{\thesubsection}{\Alph{section}.\arabic{subsection}} 

\begin{center}
\begin{Large}
\textbf{Online Appendix} \\ ~ \\
\end{Large}
\begin{LARGE}
Can Pollution Markets Work in Developing Countries? Experimental Evidence from India
\end{LARGE}
\end{center}

\begin{center}
\begin{large}
\author{Michael Greenstone, Rohini Pande, Nicholas Ryan and Anant Sudarshan}
\end{large}
\end{center}

\vspace{0.5cm}
\doublespacing


\vspace{1cm}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Experimental Design \label{sec:appendixMarket}}
	
\setcounter{table}{0}
\setcounter{figure}{0}


% Table A1: Intervention timeline 
\begin{table}[h!]
	\centering 
	\begin{threeparttable}
		\caption{Intervention timeline \label{tab:timeline}}
        \input{ExhibitsGeneratedByHand/Table_A1.tex}
		\begin{tablenotes}[flushleft]
			\setlength\labelsep{0pt}
			\item \footnotesize Compliance periods were of heterogeneous length, though most lasted approximately one month; of particular note, Period-III began in the middle of November and lasted 45 days until early January. Baseline and endline surveys collected data on plant and boiler house costs, revenue, and emissions abatement mechanisms. While CEMS device readings were collected from April 2019 onward, data availability was low until the emissions trading scheme commenced in July 2019. During mock periods, plants simulated live period transactions with monetary vouchers. We had two interregnum periods where the market was closed: the first wave of the COVID-19 pandemic and shutdowns, and Diwali in 2020. Plant production remained sufficiently high during Diwali in 2019 to continue market operations.
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\clearpage

% Table A2: Compliance periods and market caps
\begin{landscape}		
	\begin{table}[h!]
		\centering
		\begin{threeparttable}
		\caption{Compliance periods and market caps \label{tab:marketCap}}
		\input{ExhibitsGeneratedByHand/Table_A2.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item This table reports the start and end date of compliance periods and the market cap of each period. The market cap is the total amount of PM emissions – summed up across all market participants - that is allowed \textit{per month (30 days)} under the Emissions Trading scheme. The total market cap varies across compliance periods, due to the duration of the compliance period. Specifically, the total market cap in a compliance period is the market cap $\times$ 30 / (number of days in the compliance period). The per-plant cap is calculated by dividing the market cap by 162, the number of in-sample plants in the treatment arm. The market was closed during Interregnum-I due to the COVID-19 pandemic and during Interregnum-II following the Divali festival.
			\end{tablenotes}
		\end{threeparttable}
	\end{table}
	\end{landscape}

% Table A3: Sample restrictions and attrition by treatment status
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Sample determination and attrition by treatment status}
		\label{tab:sample}
		\input{ExhibitsGeneratedByHand/Table_A3.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item This table reports the sample determination and attrition during the ETS experiment. Of the original ETS-CEMS sample of 373 plants, 342 operational plants received treatment assignment in May 2019 (row 1). Of these 342 plants included in the ETS treatment randomization, 20 plants were extinct or permanently closed (row 2). The permanent shutdown status of these 20 plants has been verified with Ringelmann survey panel data covering the sample from March 2018 to June 2019, as well as regulatory inspection and audit documentation on the GPCB administrative portal. The 342 plants that received treatment assignment, less the 20 plants that received assignment while extinct or shutdown, yield 322 operational plants with treatment assignment at baseline (row 3). Four of these 322 operational-at-baseline plants were officially removed from the ETS sample by GPCB after the treatment assignment (row 4). Three of the removed plants (2 in control, 1 in treatment) are seasonal sugar cooperatives, operational for only four months of the year; the fourth treatment plant is a particle-board producing plant which uses bagasse, rather than coal, as fuel. Of the 318 in-sample plants, 13 are known to have been incompletely treated by the intervention, due to temporary ﬁnancial closure before or after the treatment assignment was done (row 6). The 304 plants surveyed at baseline are distinct from the 304 plants manually sampled, and are therefore reported separately (rows 8, 9). This paper reports experimental results from the sample of 292 plants reported at least one day of CEMS data from April 16, 2019 to April 3rd, 2021 (row 11). Of the 162 in-sample plants in the treatment group, 153 plants have market trading data (row 13).
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\clearpage 

% Table A4: Balance of plant characteristics by treatment status
% /01Code/emissions/make_balance_table_new.do
\singlespacing
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{ThreePartTable}
    \begin{TableNotes}[flushleft]
        \setlength\labelsep{0pt}
        \item \footnotesize \singlespacing This table shows differences in plant measures (panel A), plant abatement and investment cost (panel B), and plant pollution (panel C) between the treatment and control groups of plants in the baseline survey conducted from December 2018 to January 2019. This sample consists of 318 plants in the ETS experiment. In panel B, cyclone, bag filter, scrubber, and electrostatic precipitator (ESP) are different devices used to reduce emissions. Some plants did not respond to some questions in the survey. For the control group, the numbers of observations are 137 for boiler house capital expenditure, 141 for gross sales revenue, 148 for the Ringelmann score, 156 for plant total heat output, and 147 for the rest. For the treatment group, the numbers of observations are 147 for boiler house capital expenditure, 150 for gross sales revenue, 160 for Ringelmann score, 162 for plant total heat output and the number of stacks, and 157 for the rest. The first and second columns show means with standard deviations given in brackets. The third column shows the coefficient from regressions of each variable on treatment, with robust standard errors in parentheses. $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
    \end{TableNotes}
    \input{../03Output/tables/Table_A4.tex}
\end{ThreePartTable}	
\clearpage
\doublespacing



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Data \label{sec:appendixData}}
\setcounter{table}{0}
\setcounter{figure}{0}

% Figure B1: 
% 01Code/emissions/plot_emissions_distribution.do
\begin{figure}[h]
    \centering 
    \caption{Distribution of Pollution Before the Experiment\label{fig:pollutionHistograms}}
    \includegraphics[width=0.90\linewidth]{../03Output/figures/Figure_B1.pdf}
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize This figure shows the distributions of the plant PM concentration by treatment status as measured by manual iso-kinetic stack sampling at the ETS baseline (December 2018 to January 2019). One PM sample was collected from each industrial stack by a third-party laboratory. The histograms are truncated at the 95th percentile (520 mg/Nm$^3$). The red, vertical lines indicate the regulatory concentration standard of 150 mg/Nm$^3$. At the ETS baseline, 28\% of sampled plants in the control group and 34\% of sampled plants in the treatment group had readings above this standard.} \\
    \end{tabular*}
\end{figure}

% Table B1:
% Code: 01Code/trading/compute_summary_statistics.do
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Trading data summary statistics\label{tab:tradingSummary}}
		\label{tab:summ_trade}
		\begin{tabular}{@{\extracolsep{2pt}}lD{.}{.}{-1}D{.}{.}{-1}D{.}{.}{-1}} \\[-1.8ex]
			\toprule \\[-1.8ex]  
			 & \multicolumn{1}{ >{\centering\arraybackslash}m{2cm} }{All} & \multicolumn{1}{ >{\centering\arraybackslash}m{2cm} }{Purchase} & \multicolumn{1}{ >{\centering\arraybackslash}m{2cm} }{Sale}  \\ 
				 \\[-1.8ex] \midrule \\[-1.8ex]
			\multicolumn{4}{c}{\textit{Panel A: Order}} \\ 
			\input{../03Output/tables/Table_B1_A.tex} \\
			\multicolumn{4}{c}{\textit{Panel B: Trade}} \\ 
			\input{../03Output/tables/Table_B1_B} \\
			\bottomrule \\[-1.8ex]
			\end{tabular}		  
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item This table shows the mean of order quantity and price (panel A) and trade quantity and price (panel B), with the standard deviation given in the brackets.
		\end{tablenotes}
	\end{threeparttable}
\end{table}

% Figure B2: Distribution of number of bids placed per plant by compliance period
% Code: 01Code/trading/plot_histogram_number_bids.do
\begin{figure}[h]
	\centering \caption{Distribution of number of bids placed per plant by compliance period \label{fig:hist_num_bids_period}}
	\includegraphics[width=0.75\linewidth, trim=0.5cm 0.5cm 0.5cm 0.5cm, clip]{../03Output/figures/Figure_B2.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize This figure presents the distributions of number of bids placed per plant by compliance period, truncated at 40 (about 99th percentile). The bin width is 1. The red line indicates the median number of bids placed.}\\
	\end{tabular*}
\end{figure}
\clearpage


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Pollution Monitoring \label{sec:appendixCEMS}}
\setcounter{table}{0}
\setcounter{figure}{0}

% Figure C1: 
% Code: 01Code/emissions/plot_intra_day_data_availability.R
\begin{figure}[h!]
	\centering \caption{Share of data available within month for months with partial data\label{fig:intra_imputereporting}}
	\includegraphics[width=\linewidth]{../03Output/figures/Figure_C1.pdf}
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize This figure plots a histogram of data availability at the stack-month level. The only stack-months included are those which require intra-month imputation, which are those with some, but not complete, minute-level reporting throughout the month. This represents 73\% of the stack-months in the sample. The y-axis represents the portion of all plant-months in that panel which fall into the corresponding bin. Each panel represents a different quarter-year of the sample, excluding interregnum periods.}\\
	\end{tabular*}
\end{figure}

% Figure C2: Data availability from CEMS by treatment status
% Code: 01Code/emissions/plot_emissions_time_series.do
\begin{figure}[h]
    \centering 
    \caption{Data availability from CEMS by treatment status \label{fig:cemsDA}}
    \includegraphics[width=1.0\linewidth]{../03Output/figures/Figure_C2.pdf} 
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows the percentage of plants reporting, at weekly frequency, from April 2019 to March 2021. The missing pollution readings are imputed within a stack-week, but not across stacks or weeks. This sample consists of 292 plants that had at least one day of PM data from CEMS devices during the ETS experiment. The treatment group is represented by the solid (blue) line, and control group by the dashed (grey) line. The grey regions mark the ten compliance periods in the emissions market. The light blue regions mark the two interregnum periods when the emissions market was closed.}\\
    \end{tabular*}
\end{figure}
\clearpage

% Figure C3: PM emissions by treatment status
% 01Code/emissions/plot_emissions_time_series.do
\begin{figure}[h!]
    \centering 
    \caption{PM emissions by treatment status \label{fig:pollutionTimeSeries_appendix}}
    \subfiguretopcaptrue
    \subfigure[Rule A: Stack-Experiment]{\includegraphics[width=1\linewidth]{../03Output/figures/Figure_C3_A.pdf}}
    \subfigure[Rule B: Treatment-Month]{\includegraphics[width=1\linewidth]{../03Output/figures/Figure_C3_B.pdf}}
    \begin{tabular*}{1.0\textwidth}{c}
        \multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows the weekly mean per-plant PM emissions in kilograms calculated at a monthly rate equivalent, from April 2019 to March 2021. In the top panel, the missing pollution readings are imputed within stack-week, and then within stack-experiment. In the bottom panel, they are imputed within stack-week, and then within treatment-month. Appendix C.1 provides a detailed note on the construction of the PM emission variable. This sample consists of 292 plants that had at least one day of PM data from CEMS devices during the ETS experiment.  The treatment group is represented by the solid (blue) line, control group by the dashed (grey) line. The grey regions mark the ten compliance periods in the emissions market. The light blue regions mark interregnum periods when the emissions market was closed. The horizontal (red) lines denote the per-plant month market cap for each period. The aggregate market caps for each compliance period were: 280 tons per 30 days (for Mock-I, Mock-II, and Period-I), 200 tons per 30 days  (for Period-II), 180 tons per 30 days (for Period-III), and 170 tons per 30 days thereafter.}\\
    \end{tabular*}
\end{figure}
\clearpage

% Table C1: Mean of the log(PM emissions) by imputation rules
% Code: 01Code/emissions/compare_imputation_rules.do
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Mean of the log(PM emissions) by imputation rules}\label{tab:CompareImpRules}
		\input{../03Output/tables/Table_C1.tex}

		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item The table shows the mean ln[PM emissions (kg/month)] with the number of observations given in the brackets by different imputation rules in the control group, treatment group, and the whole sample. Observational unit is stack-month, excluding interregnum and mock trading periods, across 292 potential plants in the whole sample.
		\end{tablenotes}
	\end{threeparttable}
\end{table}

% Figure C4: Kernel density of PM emissions by treatment status
% Code: 01Code/emissions/compare_imputation_rules.do
\begin{figure}[h!]
	\centering \caption{Kernel density of PM emissions by treatment status\label{fig:pollutionKernels}}
\subfiguretopcaptrue

    \subfigure[PM emissions]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_C4_A_1.png}\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_C4_A_2.png}}

    \subfigure[log(PM emissions)]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_C4_B_1.png}\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_C4_B_2.png}}
\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize This figure plots the kernel density of PM emissions (kg/month) in Panel A and log(PM emissions) in Panel B, both by treatment status, in different stages of imputation described in Section C.2. Stack-Week corresponds to the emissions variable after step 2. Stack-Month, Stack-Experiment, and Treatment-Month correspond to the variables constructed based on the No Imputation Rule, the Imputation Rule A, and the Imputation Rule B, respectively. Imputing the treatment group mean causes values to converge to the group mean, so the distribution of PM emissions and that of log(PM emissions) should have less dispersion under Rule B. Since the distribution of emissions is highly positive-skewed, the emissions of most plants are less than the group mean. Rule B, therefore, inflates the emissions of those plants. As a result, the peak of the kernel density curve under Treatment-Month for the control group shifts to the right. As the distribution of PM emissions is more clustered near the mean under Rule B, the mean of log(PM emissions) should be closer to the log of mean PM emissions for Rule B. By the concavity of log function, the log of mean is no less than the mean of log values. Hence, the mean of log(PM emissions) should be higher for Rule B than others.}\\
	\end{tabular*}
\end{figure}

\clearpage

% Table C2: Imputation Rules for Missing CEMS Data
\begin{table}[h!]
	\caption{Imputation Rules for Missing CEMS Data\label{tab:impMarket}} 
	\begin{threeparttable}
	\input{ExhibitsGeneratedByHand/Table_C2.tex}
	\end{threeparttable}
\end{table}

% Table C3
% Code: 01Code/emissions/make_effect_of_cems_table.do
\begin{table}[h!]\centering
	\begin{threeparttable}
		\caption{Effects of CEMS Installation on Plant Emissions \label{tab:cems_appendix}}
	
		\input{../03Output/tables/Table_C3}
	\begin{tablenotes}[flushleft]
		 \footnotesize
				\setlength\labelsep{0pt}
	   \item Dependent variable emission measures are from GPCB's regularly scheduled manual samplings. Unit of observation is plant. Sample is restricted to plants in Phases 1 and 3 of CEMS rollout. The treatment indicator in the regression is set to the interaction of the plant being in Phase 1 and not being in the control group. The treatment indicator is therefore set to 0 for the union of all experiment control plants and all Phase 3 plants. Standard errors are clustered at the plant-level.\\
	\sym{*} \(p<0.10\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)\\
	\end{tablenotes}
	\end{threeparttable}
	\end{table}

% Table C4
% Code: 01Code/emissions/make_effect_of_cems_table.do
\newcommand{\STAB}[1]{\begin{tabular}{@{}c@{}}#1\end{tabular}}
\begin{table}[!ht]
	\centering
	\begin{threeparttable}
		\caption{Treatment effect on PM emissions (log(PM mass/month)) with different imputation rules for the control and treatment groups}
		\input{../03Output/tables/Table_C4}
		\label{tab:pollutionITT_by_imp_rule}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
            \item This table reports estimated treatment effects on PM emissions, as in Table 3, column (5) of the main text, using different imputation rules for the treatment and control groups. The outcome variable is the log of plant-level PM mass (kg) per month. A detailed note on the construction of the outcome variable is in Appendix C.1. For each cell, the row describes the imputation rule used for treated plants and the column the imputation rule used for control plants. Rule A is stack-experiment imputation. Under this rule, missing values of a stack’s daily PM mass rate are imputed using the stack’s mean PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). p70 imputes missing values of a stack’s daily PM mass rate using the stack’s 70$^{\text{th}}$ percentile of PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). p80 and p90 are identical to the p70 imputation rule except that they use the 80$^{\text{th}}$ and 90$^{\text{th}}$ percentiles of PM mass rate respectively. Market is the market imputation rule described in Table~\ref{tab:impMarket}. All regressions control for plant characteristics including capital expenditure, operating cost, log(total heat output), mean boiler installation year, and their corresponding indicators for missing values. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\clearpage

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Model specification and abatement costs\label{sec:specification}}
\label{sec:appendixModel}
\setcounter{table}{0}
\setcounter{figure}{0}


% Table D1: Variable abatement costs under alternative regulatory regimes
% 01Code/model/matlab/exhibits/tabulateCounterResults.m
\begin{landscape}
\renewcommand{\tabcolsep}{4pt}
\begin{table}[ht]
	\centering
 \begin{threeparttable}
		\caption{Variable abatement costs under alternative regulatory regimes\label{tab:counterTable2Panel}}
			\begin{tabular}{lcccccc}
				\toprule
& \multicolumn{3}{c}{Emissions = 170 tons} & \multicolumn{3}{c}{Emissions = 240 tons}\\
\cmidrule(lr){2-4} \cmidrule(lr){5-7}					&\multicolumn{1}{c}{Price}&\multicolumn{1}{c}{Cost}&\multicolumn{1}{c}{$\Delta$Cost}&\multicolumn{1}{c}{Price}&\multicolumn{1}{c}{Cost}&\multicolumn{1}{c}{$\Delta$Cost}\\
					&\multicolumn{1}{c}{(INR/kg)}&\multicolumn{1}{c}{(INR m)}&\multicolumn{1}{c}{(\%)}&\multicolumn{1}{c}{(INR/kg)}&\multicolumn{1}{c}{(INR m)}&\multicolumn{1}{c}{(\%)}\\
					&\multicolumn{1}{c}{(1)}&\multicolumn{1}{c}{(2)}&\multicolumn{1}{c}{(3)}&\multicolumn{1}{c}{(4)}&\multicolumn{1}{c}{(5)}&\multicolumn{1}{c}{(6)}\\
					\midrule

    \multicolumn{7}{c}{\textit{Panel A: Iso-Elastic MAC Curve}}\\
	\input{../03Output/tables/Table_D1_A.tex} \\
    \multicolumn{7}{c}{\textit{Panel B: Step-Function MAC Curve}}\\
	\input{../03Output/tables/Table_D1_B.tex} \\
     \bottomrule
     \end{tabular}
   \begin{tablenotes}[flushleft]
     \footnotesize
			\setlength\labelsep{0pt}
   \item The table shows the results of counterfactual simulations under different regulatory regimes. Each row represents a different regime. Each panel corresponds to a different functional form assumption on the plant level marginal abatement cost curve. Within each panle the first row is the emissions market. The second through final rows in each panel are different command and control regimes that vary in how the emissions target is set for each plant. Constant emissions rate sets a single fixed ratio of emissions to heat output capacity for all plants. Constant emissions rate with error allows for idiosyncratic variation in the constant rate across plants. Capacity-based rate sets an emissions rate as a function of plant capacity, such that larger plants can have higher or lower rates of emission per unit capacity. Capacity-based rate with error allows for the capacity-based rate to idiosyncratically vary across plants. Finally, capacity-based rate with correlated error is the same as capacity-based rate with error except that the idiosyncratic error is drawn with a negative -0.1 correlation with estimated plant marginal abatement cost shocks. Columns 1 to 3 show results for emissions of 170 tons per month (the treatment level) and columns 4 to 6 for emissions of 240 tons per month (the control level). Within each set of three columns the variables show the market price (if applicable), the total variable abatement costs per month, and the change in abatement costs relative to the emissions market.
			\end{tablenotes}
	\end{threeparttable}
\end{table}
\end{landscape}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Marginal Cost Curve Specification \label{sec:appendixMAC}}
\setcounter{table}{0}
\setcounter{figure}{0}

% Figure E1: 
\begin{figure}[h]
    \centering
    \caption{Example Step-Function MAC Curve for Plant $i$ in Period $t$}
    \label{fig:example_mac_tikz}
    \centering
    \input{ExhibitsGeneratedByHand/Figure_E1.tex}
\end{figure}


% Figure E2:
% model/code/exhibits/plotFitTimeSeriesPrice.m
\begin{figure}[h]
	\centering \caption{Model Fit to Market-Clearing Prices with Step-Function Alternative\label{fig:modelFitStep}}
	\includegraphics[width=0.8\linewidth]{../03Output/figures/Figure_E2.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows the fit of the step-function and iso-elastic MAC models compared to the time series of market and bid prices by compliance period. The solid red line with square points is the time series of market-clearing prices in the fitted model with step-function MACs. The solid blue line with circular points is the time series of market-clearing prices in the fitted model with the original iso-elastic MACs. The models are fit based on bids in the first half of each compliance period. The dashed (black) line is the time series of mean bid prices in the data and the dotted (black) line is the time series of market-clearing prices.}
	\end{tabular*}
\end{figure}

% Figure E3: 
% Code: /01Code/model/matlab/exhibits/plotFitHistEmissions.m
\begin{figure}[h]
	\centering \caption{Histograms of predicted versus observed emissions \label{fig:histEmissionsStep}}
\subfiguretopcaptrue

    \subfigure[Step-function MAC period 4]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_A.pdf}}
    \subfigure[Step-function MAC period 8]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_B.pdf}} \\
    
    \subfigure[Iso-elastic MAC period 4]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_C.pdf}}
    \subfigure[Iso-elastic MAC period 8]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_D.pdf}} \\
    
    \subfigure[Observed data period 4]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_E.pdf}}
    \subfigure[Observed data period 8]{\includegraphics[width=0.49\linewidth]{../03Output/figures/Figure_E3_F.pdf}}\\
    
	\begin{tabular*}{1.0\textwidth}{c}
	 	\multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows predicted and observed emissions levels in 2 periods for 2 different MAC curve specifications. Panels A and B show predicted emissions when running our emissions market model under a step-function MAC curve in periods 4 and 8 respectively. Panels C and D show the same except using the iso-elastic MAC curve. Panels E and F show the observed distribution of emissions in those periods.} \\
	 \end{tabular*}
\end{figure}

% Table E1: Counterfactual results with heterogeneous MAC elasticity
% model/matlab/exhibits/tabulateCounterResults.m
\begin{landscape}
    \begin{table}[ht]
		\centering
	 \begin{threeparttable}
			\caption{Variable abatement costs under alternative regulatory regimes (with Heterogeneity by APCD)\label{tab:counterTableHetero}}
				\input{../03Output/tables/Table_E1.tex}
	   \begin{tablenotes}[flushleft]
		 \footnotesize
				\setlength\labelsep{0pt}
	   \item The table shows the results of counterfactual simulations under different regulatory regimes. Each row represents a different regime. The first row is the emissions market. The second through final rows are different command and control regimes that vary in how the emissions target is set for each plant. Constant emissions rate sets a single fixed ratio of emissions to heat output capacity for all plants. Constant emissions rate with error allows for idiosyncratic variation in the constant rate across plants. Capacity-based rate sets an emissions rate as a function of plant capacity, such that larger plants can have higher or lower rates of emission per unit capacity. Capacity-based rate with error allows for the capacity-based rate to idiosyncratically vary across plants. Finally, capacity-based rate with correlated error is the same as capacity-based rate with error except that the idiosyncratic error is drawn with a negative -0.1 correlation with estimated plant marginal abatement cost shocks. Columns 1 to 3 show results for emissions of 170 tons per month (the treatment level) and columns 4 to 6 for emissions of 240 tons per month (the control level). Within each set of three columns the variables show the market price (if applicable), the total variable abatement costs per month, and the change in abatement costs relative to the emissions market.
				\end{tablenotes}
		\end{threeparttable}
	\end{table}
\end{landscape}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\clearpage
\section{Appendix: Additional Results \label{sec:appendixResults}}
\setcounter{table}{0}
\setcounter{figure}{0}

% Table F1: Engineering estimates of abatement costs under ideal operating efficiency, if a cyclone is already operating
% Code: 01Code/phone_survey/create_cost_tables.do
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Engineering estimates of abatement costs under ideal operating efficiency, if a cyclone is already operating}
		\label{tab:apcdcosts_cyclone}
		\input{../03Output/tables/Table_F1.tex}
        \begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item \textit{Note.} Table displays engineering estimates of abatement cost for different APCDs and boiler capacities. We assume one cyclone is already operating when calculating the quantity of abatement, and we assume each APCD is purchased in isolation. Costs can be compared with those in other tables at a rate of INR 70 to USD 1. Capital costs are amortized to a monthly flow value. All plants are assumed to have a raw inlet concentration of 2,000 mg/Nm\textsuperscript{3}; in practice it can vary between 1,000 mg/Nm\textsuperscript{3} and 10,000 mg/Nm\textsuperscript{3}. This is converted to a monthly mass rate via a volumetric flow rate collected at baseline, assuming continuous operation for 16 hours/day and 25 days/month. Of plants with boilers in our analysis sample, the boiler capacity (BC) distribution is: 11\% have 2-3 TPH BC, 47\% have 4-7 TPH BC, 36\% have 8-14 TPH BC, 6\% have 15+ TPH BC.
		 \end{tablenotes}
		% \begin{tablenotes}[flushleft]
		% 	\footnotesize
		% 	\setlength\labelsep{0pt}
		% 	\item The table is the same as Table~\ref{tab:apcdcosts} except one cyclone is already assumed to be operating when calculating the quantity of abatement. 
		% \end{tablenotes}
	\end{threeparttable}
\end{table}

% Table F2:
% Code: 01Code/phone_survey/create_regression_tables_KW.do
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Treatment effects on the presence of abatement devices\label{tab:APCDpresence}}
		\input{../03Output/tables/Table_F2.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item This table reports the effects of treatment assignment on the presence of APCDs. All specifications control for the corresponding baseline value. Robust standard errors are given in parentheses with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
		\end{tablenotes}
	\end{threeparttable}
\end{table}

% Figure F1: 
% Code: 01Code/trading/make_elasticity_coefplot.do
\begin{figure}[h!]
	\centering 
    \caption{Elasticity estimate by weeks remaining in the order period \label{fig:elasticity_coefplot}}
	\includegraphics[width=0.75\linewidth, trim=0.5cm 0.5cm 0.5cm 0.5cm, clip]{../03Output/figures/Figure_F1.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize The top panel presents the coefficients of log(emissions as bid) from regressing ln(bid price) on log(emissions as bid) and plant $\times$ period fixed effects, estimated with different sample truncations defined by the number of weeks remaining in the order period. The bottom panel shows the number of bids placed in different sample truncations.}\\
	\end{tabular*}
\end{figure}
\clearpage

% Table F3: 
% Code: 01Code/emissions/make_balance_table_reporting_split.do
\singlespacing
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{ThreePartTable}
	\begin{TableNotes}[flushleft]
		\setlength\labelsep{0pt}
		\item \footnotesize \singlespacing This table shows differences in plant scale (panel A), plant abatement and investment costs (panel B), and plant pollution (panel C) between the plants who report above and below the median level across plants. Each plants level of reporting is calculated as the average minute-level CEMS data availability across the full sample period and across all stacks belonging to that plant. The only plants which are included in this table are those in the analysis set. This sample consists of 292 plants that had at least one day of PM data from CEMS devices during the ETS experiment. In panel B, cyclone, bag filter, scrubber, and electrostatic precipitator (ESP) are different air pollution control devices (APCDs). Some plants did not respond to some questions in the survey and so certain variable rows have fewer observations than the full sample size. The first and second columns show means with standard deviations given in brackets. The third column shows the coefficients from regressions of each variable on treatment, with robust standard errors in parentheses. $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
	\end{TableNotes}
	\input{../03Output/tables/Table_F3.tex}
\end{ThreePartTable}

% Table F4: 
% Code: 01Code/emissions/make_missingdata_abatecost_reg.R
\doublespacing
\begin{table}[h!]
	\centering
	\begin{threeparttable}
		\caption{Treatment effect on reporting by predicted plant abatement costs}
		\label{tab:missingdata_abatecost}
		\small
		\input{../03Output/tables/Table_F4.tex}
		\begin{tablenotes}[flushleft]
			\footnotesize
			\setlength\labelsep{0pt}
			\item Unit of observation is plant. Predicted Abatement Cost variable for industry set to the engineering estimate of the average abatement cost per kg from Table~\ref{tab:apcdcosts_cyclone} assuming Boiler Capacity = 8TPH) for the most advanced abatement technology of the plant: Cyclone less advanced than bag-filter less advanced than scrubber less advanced than ESP. Share of day not reporting calculated at industry level as average over industry's stacks and over all days (excluding interregnum). Robust standard errors in parentheses.
		\end{tablenotes}
	\end{threeparttable}
\end{table}
\clearpage

% Table F5:
% Code: 01Code/emissions/make_regression_table.do
\begin{landscape}

	\begin{table}[!ht]
		\centering
		\begin{adjustbox}{max width=1.4\textwidth}
		\setlength{\tabcolsep}{2pt}
		\begin{threeparttable}
			\caption{Treatment effects on PM emissions ($\log$(PM mass/month)) controlling for data availability}
			\input{../03Output/tables/Table_F5.tex}
		  	\label{tab:pollutionITT_DA}
			\begin{tablenotes}[flushleft]
				\footnotesize
				\setlength\labelsep{0pt}
				\item This table reports the estimated treatment effects on PM emissions adding average availability. The outcome variable is the log of plant-level PM mass (kg) per month. A detailed note on the construction of the outcome variable is in Appendix C.1. Columns 5 and 6 impute data with Imputation Rule A: \textit{Stack-Experiment}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the stack's mean PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). Columns 7 and 8 impute data with Imputation Rule B: \textit{Treatment-Month}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the monthly mean PM mass rate of the stack's treatment group. All columns control for plant characteristics including capital expenditure, operating cost, log(total heat output), mean boiler installation year, and their corresponding indicators for missing values. In addition to plant controls, columns 2, 4, 6, and 8 add year-month fixed effects to control for time variant differences common in each plant. We also apply the inverse probability weighting method in columns 3 and 4. The probability of reporting in a month is predicted using a probit model where the only explanatory variable is an indicator variable for the treatment status in a prior experiment that randomized CEMS installation timing. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
			\end{tablenotes}
		\end{threeparttable}
	\end{adjustbox}
	\end{table}
\end{landscape}
\clearpage

% Figure F2:
% 01Code/trading/covid_net_demand_graphs.R 
\begin{figure}[h]
	\centering \caption{Net demand before and after COVID\label{fig:covid_net_demand_scatter}}
	\includegraphics[width=0.7\linewidth]{../03Output/figures/Figure_F2.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize The figure shows net demand (emissions -- initial allocation) for each industry averaged across pre vs post COVID compliance periods. Each point represents a single industry. The solid black line is the OLS fit for the data The dotted black line is the $y=x$ line. We omit the 9 industries with values of magnitude greater than 2000 for ease of visualization.} \\
	\end{tabular*}
\end{figure}

% Table F6: 
% Code: 01Code/emissions/make_regression_table_covid_inter.do
\begin{landscape}
	\begin{table}[!ht]
		\centering
		\begin{adjustbox}{max width=1.6\textwidth}
		\setlength{\tabcolsep}{2pt}
		\begin{threeparttable}
			\caption{Treatment effects on PM emissions ($\log$(PM mass/month)) before and after COVID}
		  	\input{../03Output/tables/Table_F6}
		  	\label{tab:pollutionITT_Covid}
			\begin{tablenotes}[flushleft]
				\footnotesize
				\setlength\labelsep{0pt}
				\item This table reports the estimated treatment effects on PM emissions. The outcome variable is the log of plant-level PM mass (kg) per month. A detailed note on the construction of the outcome variable is in Appendix . Post-Covid is defined as periods 7 to 10. Columns 5 and 6 impute data with Imputation Rule A: \textit{Stack-Experiment}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the stack's mean PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). Columns 7 and 8 impute data with Imputation Rule B: \textit{Treatment-Month}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the monthly mean PM mass rate of the stack's treatment group. Post-COVID coded as after March 2020. All columns control for plant characteristics including capital expenditure, operating cost, log(total heat output), mean boiler installation year, and their corresponding indicators for missing values. In addition to plant controls, columns 2, 4, 6, and 8 add year-month fixed effects to control for time variant differences common in each plant. We also apply the inverse probability weighting method in columns 3 and 4. The probability of reporting in a month is predicted using a probit model where the only explanatory variable is an indicator variable for the treatment status in a prior experiment that randomized CEMS installation timing. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
			\end{tablenotes}
		\end{threeparttable}
	\end{adjustbox}
	\end{table}
\end{landscape}
\clearpage

% Table F7:
%|model/matlab/exhibits/tabulateCounterResults.m
\begin{landscape}
    \begin{table}[ht]
		\centering
	 \begin{threeparttable}
			\caption{Variable abatement costs under alternative regulatory regimes using only pre-COVID data\label{tab:counterPrecovidComp}}
				\begin{tabular}{lcccccc}
					\toprule
	& \multicolumn{3}{c}{Emissions = 170 tons} & \multicolumn{3}{c}{Emissions = 240 tons}\\
	\cmidrule(lr){2-4} \cmidrule(lr){5-7}					&\multicolumn{1}{c}{Price}&\multicolumn{1}{c}{Cost}&\multicolumn{1}{c}{$\Delta$Cost}&\multicolumn{1}{c}{Price}&\multicolumn{1}{c}{Cost}&\multicolumn{1}{c}{$\Delta$Cost}\\
						&\multicolumn{1}{c}{(INR/kg)}&\multicolumn{1}{c}{(INR m)}&\multicolumn{1}{c}{(\%)}&\multicolumn{1}{c}{(INR/kg)}&\multicolumn{1}{c}{(INR m)}&\multicolumn{1}{c}{(\%)}\\
						&\multicolumn{1}{c}{(1)}&\multicolumn{1}{c}{(2)}&\multicolumn{1}{c}{(3)}&\multicolumn{1}{c}{(4)}&\multicolumn{1}{c}{(5)}&\multicolumn{1}{c}{(6)}\\
						\midrule
	
		\multicolumn{7}{c}{\textit{Panel A: Iso-Elastic MAC Curve}}\\
		\input{../03Output/tables/Table_F7_A.tex} \\
						
		\multicolumn{7}{c}{\textit{Panel B: Step-Function MAC Curve}}\\
		\input{../03Output/tables/Table_F7_B.tex} \\
	
		 \bottomrule
		 \end{tabular}
		 \begin{tablenotes}[flushleft]
		 \footnotesize
				\setlength\labelsep{0pt}
	   \item The table shows the results of counterfactual simulations under different regulatory regimes. Each row represents a different regime. The first row is the emissions market. The second through final rows are different command and control regimes that vary in how the emissions target is set for each plant. Constant emissions rate sets a single fixed ratio of emissions to heat output capacity for all plants. Constant emissions rate with error allows for idiosyncratic variation in the constant rate across plants. Capacity-based rate sets an emissions rate as a function of plant capacity, such that larger plants can have higher or lower rates of emission per unit capacity. Capacity-based rate with error allows for the capacity-based rate to idiosyncratically vary across plants. Finally, capacity-based rate with correlated error is the same as capacity-based rate with error except that the idiosyncratic error is drawn with a negative -0.1 correlation with estimated plant marginal abatement cost shocks. Columns 1 to 3 show results for emissions of 170 tons per month (the treatment level) and columns 4 to 6 for emissions of 240 tons per month (the control level). Within each set of three columns the variables show the market price (if applicable), the total variable abatement costs per month, and the change in abatement costs relative to the emissions market. Data used for estimation is restricted to pre-COVID periods (periods 1 to 6) only.
				\end{tablenotes}
		\end{threeparttable}
	 \end{table}
\end{landscape}

% Figure F3: 
% Code: 01Code/trading/plot_histogram_permit_consumption.do
\begin{figure}[h]
	\centering \caption{Distribution of Emissions over Final Permit Holdings by Compliance Period without GPCB’s Period 7 Adjustment \label{fig:cems_sampling_concur}}
	\includegraphics[width=0.7\linewidth ]{../03Output/figures/Figure_F3.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize This figure plots the distributions of (emissions / final permit holdings $\times$ 100\%) across treated plants (N = 156) by compliance period. Final permit holdings are the total number of permits a plant held at the end of the true-up period after each compliance period. Emissions data and permit holdings are from the administrative records of the market operator. Permit holdings are adjusted to remove those granted in GPCB's period 7 adjustment. Emissions are the validated emissions for each plant, which include any imputed emissions filled-in for periods of missing data. These validated emissions are used to determine compliance.} \\
	\end{tabular*}
\end{figure}

% Figure F4: 
% Code: 01Code/emissions/plot_cems_manualsampling.R
\begin{figure}[h]
	\centering \caption{Simulatenous CEMS and sampling comparison \label{fig:cems_sampling_concur}}
	\includegraphics[width=0.7\linewidth ]{../03Output/figures/Figure_F4.pdf} 
	\begin{tabular*}{1.0\textwidth}{c}
		\multicolumn{1}{p{.98\hsize}}{\footnotesize The figure plots CEMS readings against concurrent manual samplings. Unit of observation is an industry. The solid black line is the OLS fit for the data The dotted black line is the $y=x$ line. We restrict the graph to only those CEMS readings with at least 15\% data availability during the appropriate window, and to those with concentrations less than 2000 mg/Nm3.} \\
	\end{tabular*}
\end{figure}

% Table F8: 
% Code: 01Code/emissions/make_regression_table_device_fe.do
\begin{landscape}
	\begin{table}[!ht]
		\centering
		\begin{adjustbox}{max width=1.42\textwidth}
		\setlength{\tabcolsep}{1.7pt}
		\begin{threeparttable}
			\caption{Treatment effects on PM emissions ($\log$(PM mass/month)) depending on device type}
		  	\input{../03Output/tables/Table_F8.tex}
		  	\label{tab:pollutionITT_device}
			\begin{tablenotes}[flushleft]
				\footnotesize
				\setlength\labelsep{0pt}
				\item This table reports the estimated treatment effects on PM emissions. The outcome variable is the log of plant-level PM mass (kg) per month. A detailed note on the construction of the outcome variable is in Appendix C.1. Columns 5 and 6 impute data with Imputation Rule A: \textit{Stack-Experiment}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the stack's mean PM mass rate across the experiment (July 2019 to March 2021, excluding interregnum). Columns 7 and 8 impute data with Imputation Rule B: \textit{Treatment-Month}. Under this rule, missing values of a stack's daily PM mass rate are imputed using the monthly mean PM mass rate of the stack's treatment group. We add fixed effects for the different types of abatement devices which an industry has across all of its stacks (Type 1, Type 2, or both). Type 1 Devices are the omitted level of device type fixed effect. Approximately 80\%, 10\%, and 10\% of plants are set to Type 1, Type 2, and having both types, respectively. All columns control for plant characteristics including capital expenditure, operating cost, log(total heat output), mean boiler installation year, and their corresponding indicators for missing values. In addition to plant controls, columns 2, 4, 6, and 8 add year-month fixed effects to control for time variant differences common in each plant. We also apply the inverse probability weighting method in columns 3 and 4. The probability of reporting in a month is predicted using a probit model where the only explanatory variable is an indicator variable for the treatment status in a prior experiment that randomized CEMS installation timing. Robust standard errors in parentheses are clustered at the plant level with statistical significance indicated by $^{*}p<$0.10; $^{**}p<$0.05; $^{***}p<$0.01.
			\end{tablenotes}
		\end{threeparttable}
	\end{adjustbox}
	\end{table}
\end{landscape}
\clearpage

% Table F9: 
% Code: 01Code/emissions/make_missing_data_analyses.R
\begin{landscape}
	\singlespacing
	
	\centering
	\footnotesize
	\input{ExhibitsGeneratedByHand/Table_F9.tex}
\end{landscape}




	
\end{document}